Feature Selection for Face Based on Adaptive Parameter Multi-Objective Optimization

针对传统多目标算法在人脸识别的特征选择问题中的搜索性能稳定性差和缓慢的收敛速度，本研究提出了一种创新的基于自适应参数的多目标优化算法(Adaptive Parameter Multi-objective Particle Swarm Optimization, APMPSO)。APMPSO通过在不同阶段合理分配全局搜索和局部搜索的力度，算法能够更高效地利用搜索资源，减少不必要的搜索过程，从而提高收敛速度和收敛精度。同时具有更好的适应性，能够根据具体的问题和数据集特点自动调整参数。通过在迭代过程中动态改变参数，算法可以更好地适应不同的搜索环境，提高在各种人脸识别问题上的性能。本算法与多种多目标优化算法在收敛速度，人脸特征选择效率问题上进行了比较，实验结果证明，APMPSO能在人脸识别问题上提供更好的帮助，显示出与粒子群算法相比的显著优势，显示出超前的应用效率。In view of the poor search performance stability and slow convergence speed of traditional multi-objective algorithms in the feature selection problem of face recognition, this study proposes an innovative multi-objective optimization algorithm based on adaptive parameters (Adaptive Parameter Multi-objective Particle Swarm Optimization, APMPSO). By reasonably allocating the strength of global search and local search at different stages, the APMPSO algorithm can more efficiently utilize search resources and reduce unnecessary search processes, thereby improving the convergence speed and convergence accuracy. At the same time, it has better adaptability and can automatically adjust parameters according to specific problems and data set characteristics. By dynamically changing parameters during the iteration process, the algorithm can better adapt to different search environments and improve performance on various face recognition problems. This algorithm is compared with a variety of multi-objective optimization algorithms in terms of convergence speed and face feature selection efficiency. The experimental results show that APMPSO can provide better help in face recognition problems, showing significant advantages over particle swarm algorithms and advanced application efficiency.